A novel method for camera calibration and image alignment of a thermal/visible image fusion system

Abstract

This paper presents a novel method for camera calibration and image alignment of an image fusion system that consists of a thermal and a color camera. The calibration board that consists of a heated metal and an acrylic plate with asymmetric circle grid pattern was developed. The board was used to calibrate intrinsic parameters of the thermal camera and also used to find the homography to align images of the two cameras. The aligned images have been fine adjusted for their misalignment due to difference in projection centers by region of interest (ROI) setting. The visible camera was calibrated separately by using a typical calibration board with an asymmetric circle grid printed paper. The evaluation was performed in two scenarios. The first one is an indoor static scene. And the second one is an outdoor dynamic scene that is a vehicle tracking application by using CAMShift algorithm. The results validate the proposed method with high accuracy.